AI-Driven Avatars in Immersive Virtual Environments: Workflow and Case Study of the Temple of Demeter, Greece
Abstract: Generative Artificial Intelligence (AI) and Virtual Reality (VR) technologies are changing education and offer new opportunities for how people interact with environments. Technological advances over the past few decades have lowered barriers to creating virtual environments; however, there are still challenges, particularly when creating realistic virtual environments of real places. Realism is important as the fidelity of virtual environments influences user experience. In addition, methods and techniques that can facilitate ease of interacting with these environments are needed to streamline the user experience. One promising way to do this is incorporating AI-driven avatars into realistic scenes, allowing users to use natural languing immersive learning experiences focused on sustainability education that incorporate realistic scage to interact with and learn about the landscapes. To this end, we have developed workflows for designenarios and natural interaction with AI avatars. This study created point cloud digital 3D models from photogrammetry and incorporated these into Unreal Engine 5. We then integrated generative AI avatars into the environment, enabling natural language interaction between users and an AI tutor. The integration facilitates interactive participation and enables high-precision digital reproduction of the physical environment. The novel proposed workflow is presented via a case study for a virtual study abroad experience in Naxos, Greece, using an AI-driven tutor to educate about the historical aspects of the island. This project provides the beneficial learning experience of study abroad experiences without the economic and environmental costs of sending students on field excursions. We recommend constructing immersive education experiences using real-world environments and natural AI-driven conversations and demonstrating its potential to revolutionize social interaction, historical heritage preservation, and sustainable pedagogy.
Key words: Generative artificial intelligence, AI-driven avatars, Virtual Reality, Heritage Preservation, Immersive Experience
Realistic virtual environments have myriad applications in a variety of disciplines including urban design and planning, virtual tourism, film and gaming, education, disaster management, and preservation and cultural heritage. However, creating virtual environments that are perceived as realistic is usually very time-consuming and expensive (KIOURT ET AL., 2017). Recent advances in hardware and software have accelerated the creation of and interaction with realistic environments, though there are still challenges. With the convergence of generative artificial intelligence (AI) and realistic digital 3D models of real-world locations, new types of immersive experiences can be created that will change how people interact with and learn about places. Here we demonstrate how new technologies can be applied to rapidly create realistic digital environments, virtual exploration, and AI-driven interactions. In this study, we demonstrate how to create a digital model of a real-world location using point cloud modeling and integrate that model with a video-game engine to create an immersive virtual reality environment. The immersive environment and simulated physical systems is used to replicate virtual roaming of a historical site. A novel approach to human-database interaction is applied by using conversational generative AI to support natural language interaction with an avatar allowing users to learn about the site. In the following sections, we provide background on 3D modeling techniques, generative AI, and related work.
Fig1. The concept of the developed workflow for AI-driven immersive learning
In recent years generating 3D models of real locations has transitioned from manually creating models using CAD software to automatically generating models from photographs. Point cloud modeling uses Structure of Motion (SfM) and Multi-View Stereo (MVS) algorithms to generate dense point clouds from photographs that represent 3D objects and structures in a detailed and accurate manner (PEPE ET AL., 2022). The SfM methodology is a widely used method in the fields of photogrammetry and computer vision that uses overlapping images to create 3D surface models. Standard consumer-grade cameras can capture images, hence rendering SfM a cost-effective tool that may be used in conjunction with other 3D technologies like terrestrial and airborne laser scanning (LIDAR). The use of unmanned aerial vehicles (UAVs) enables the acquisition of photographic datasets that possess spatial coordinates. By employing SfM and MVS algorithms, users can rapidly capture and create digital models of landscapes. Furthermore, these algorithms facilitate the categorization of features within the 3D point cloud. Using photo datasets containing spatial coordinates and overlaps proves to be highly advantageous for researchers in various disciplines such as geoinformatics, computer vision, archaeology, architectural research, human-computer interaction, algorithm evaluation for structural and multi-view stereo reconstruction in motion, point cloud processing, semantic 3D reconstruction, virtual tourism, and applications related to virtual reality and augmented reality (PEPE ET AL., 2022).
AI broadly refers to the application of computers to simulate intelligent behavior with little human guidance (KONAR, A., 2018). Generative AI, a subset of AI more generally, relies on machine learning models that have been trained using vast amounts of data (HALEEM ET AL., 2022). These models, known as mega-models, serve as the foundation for the functioning of generative AI systems. Researchers utilize machine learning techniques to facilitate the creation and advancement of cutting-edge digital generating applications. Generative AI uses a combination of manually crafted algorithms and machine learning to produce novel material by leveraging pre-existing databases (e.g., text, photos, images, music, etc.) (DWIVEDI ET AL., 2023). Generative AI techniques employ statistical models and probabilities to generate synthetic artifacts using unsupervised or partially supervised machine learning methodologies (CASTIGLIONI ET AL., 2021). Generative AI programs such as ChatGPT (https://chat.openai.com/) and Midjourney (https://www.midjourney.com/home) have gained significant popularity due to their ability to generate text and visuals.
The field of generative AI encompasses two primary categories: Generative Adversarial Networks (GANs) and Generative Pretrained Transformers (GPTs) (GOODFELLOW ET AL., 2014). The present framework of GANs consists of generator networks and discriminator networks (LI ET AL., 2017). Generator networks are responsible for generating and synthesizing data (KU & LEE, 2023). When the discriminator network successfully identifies the material generated by the generator network, it becomes unable to differentiate between synthetic and authentic content, resulting in the synthetic content being classified as genuine (GOODFELLOW ET AL., 2014; RADFORD ET AL., 2015). Large-scale language model processing systems employ machine learning algorithms to analyze user text and pre-existing text databases, generating text that resembles human language. This facilitates seamless and natural communications between humans and computers. ChatGPT and Midjourney are developed using this underlying framework to produce text that resembles human discourse and pictures that resemble human drawings. ChatGPT is an application for engaging in chat conversations that are generated from pre-existing language models, namely GPT 3.0 and GPT 4.0 (ABRAMSKI ET AL., 2023). The system possesses the ability to identify text input provided by users and generate contextually appropriate responses, resembling the comprehension of human language (BAIDOO-ANU & OWUSU ANSAH, 2023). The model has garnered significant interest across multiple disciplines, encompassing education, engineering, journalism, medicine, as well as economics and finance, among other domains (AYDIN & KARAARSLAN, 2023). ChatGPT possesses the capacity to offer tailored and efficient educational encounters through the provision of personalized feedback and explanations to students. Nevertheless, the functionality of ChatGPT is subject to several constraints, primarily its reliance on a solitary text-based dialogue interface (GRASSINI, 2023). In the current state, the linguistic output produced by ChatGPT is similar to human language, albeit with certain limitations of the accuracy of the information provided. Research has demonstrated that ChatGPT exhibits biases and inaccuracies when it comes to spreading knowledge, posing a significant obstacle in the realm of education (BISWAS, 2023).
Significant advancements have been observed in the integration of VR technology in the education domain in the past five years. One study conducted by Radianti et al. (RADIANTI ET AL., 2020) showed that immersive VR facilitated improved comprehension and educational achievements. However, the primary focus of the research was on scenarios that occur within the classroom setting. This highlights the need for more extensive and practical applications of VR in real world environments In this context a study undertaken by Walmsley et al. (WALMSLEY ET AL., 2020) investigated the creation and effectiveness of a 360 virtual reality instructional tool, yielding noteworthy outcomes. While promising, the study relied on360-image based VR environments which are static and less interacive than 3D model based VR. Haskins et al. (HASKINS ET AL., 2020) conducted a study where they demonstrated that the use of VR 3D environments increased student engagement. Nevertheless, the study conducted by the researchers failed to place significant emphasis on the advancement and use of authentic virtual worlds that accurately replicate real-world settings. Furthermore, Roumana et al. (ROUMANA ET AL., 2022) proposed an innovative approach to cultural heritage education by integrating VR technology with realistic environmental models to enhance cultural immersion. While promising, the research did not delve into specific benefits of integrating AI tools with 3D environments, which is a promising new area for teaching and learning.
The integration of generative AI in the domain of education has emerged as a potential avenue for revolutionizing teaching and learning. Promising developments are in the creation of generative models that can augment student learning and increase engagement. In a recent paper Dwivedi et al. (DWIVEDI ET AL., 2023) highlight the efficacy of generative AI in facilitating individualized learning trajectories. Jeon et al. (JEON ET AL., 2023) employed generative techniques to create educational chatbots designed for language acquisition. This innovative approach facilitated a conversational learning trajectory. Further, the use integrating AI with simulated characters for education allows for the provision of adaptive content tailored to the specific needs of learners. These AI characters can take on many roles such as playing coaches, program creators, and teachers. To this end Chheang et al. (CHHEANG ET AL., 2023) developed an intelligent tutor for anatomy education within a VR setting. However, their study did not include a dedicated database tailored to anatomy education, and the intelligent tutor was shown to be ineffective in delivering appropriate instructional content. Furthermore, the integration of the intelligent instructor with the actual environment was not accomplished, and the intelligent entity was unable to identify the things present in the environment. This highlights the substantial scope for further innovation in this field. Nevertheless, a significant deficiency in current AI education programs is the absence of a comprehensive examination of the dynamic interplay between AI characters and individuals engaged in the learning process.The current scope of AI applications is confined to structured learning models, while the potential of natural language processing applications remains largely untapped. The use of generative AI in education has a significant impact. Our research involves the integration of AI into VR and its combination with real-world 3D models. This integration aims to provide learners with an immersive and interactive educational experience that offers precise data information.
The global mass tourism industry has been expanding rapidly over the past 15 years and is associated with negative environmental impacts (AFRIN ET AL., 2013; LEKA ET AL., 2022). These impacts are evident in our case study site, Naxos Island, Greece which is undergoing gradual erosion and loss due to the combined impact of natural processes and human activities (EVELPIDOU ET AL., 2021), posing a significant threat to the preservation and education of the islands’ biodiversity and cultural heritage. The process of digitizing architectural and cultural sites holds inherent significance within the realm of heritage preservation. Simultaneously, the digital preservation of historical sites serves to enhance their accessibility, facilitating broader engagement from a diverse audience using online platforms. The accessibility of historical and cultural resources has a positive influence on the preservation of history and culture, as well as the advancement of sustainable development concepts. The majority of digital engagements with these websites are confined to passive endeavors, wherein users are restricted to clicking through and retrieving information. Regrettably, digital platforms frequently fall short of delivering engaging guided tours and interactive experiences to their visitors. To overcome these challenges this study aimed to explore the use of realistic virtual environments and AI driven avatars for increasing engagement with cultural heritage sites.
This study used a consumer-grade UAV (DJI mini 3 pro) to capture 243 aerial oblique photos of the site at angles ranging from 30 degrees to 45 degrees which were subsequently used to create a 3D model. The images were processed in ContextCapture (now iTwin Capture Modeler, https://www.bentley.com/software/itwin-capture-modeler/.) to generate an intricate mesh composed of textured triangles, using the spatial coordinates derived from the pictures. The software then goes through a process of image matching, 3D reconstruction, and seamless texture mapping, to create an authentic 3D textured model. The modeling process is an unsupervised process, and in this modeling we have a total of 243 control points, which are provided by 243 UAV aerial photos with camera angles. The 3D model was exported as an OBJ file and enhanced using Blender to rectify damaged meshes and eliminate any flaws or impurities present in the model.
After error correction the model was exported as an FBX file and imported into Unreal Engine 5 (https://www.unrealengine.com/). Unreal Engine was used due to its high level of realism and superior simulated physics when compared to other game engines. Unreal Engine also has a variety of VR templates and feature called Blueprint that enable users to implement VR gameplay mechanics without the need to code. These presets encompass guided tours, object manipulation, rotation, and many functionalities. These functionalities permit individuals to freely explore and engage with virtual models, while the incorporation of realistic lighting effects and physical systems enhances the level of immersion experienced by users.
To facilitate users interacting with AI-driven avatars we used ConvAI (https://convai.com/) ConvAI provides AI-driven avatars to augment virtual experiences by facilitating human engagement with AI in video games and virtual worlds. The process uses character modeling and animation from video games,to the creation of realistic avatars.
ConvAI provides an enhanced level of contact between humans and virtual avatars using various technological components such as Natural Language Processing (NLP), speech recognition, speech-to-text conversion, ChatGPT, and customizable databases. ConvAI provides the functionality to create custom avatars and update the avatar knowledge using a custom text based database development. ConvAI also supports avatars created using Ready Player Me (https://readyplayer.me/) which use personal photographs to construct virtual personas. The system offers both virtual characters and rudimentary animation functionalities and usesf OVRLipSync technology to improve the authenticity of an avatar's verbal communication by generating lifelike lip movements that correspond with the character's speech.
To support seamless and authentic interactions between people and AI avatars ConvAI uses advanced speech recognition technology that converts spoken language into text. This allows users to engage in interactions through voice-based input. The recognized text content is transmitted to ChatGPT, where ConvAI converts it into synthetic voice output. Simultaneously, ConvAI allows users to customize avatars by including distinct backgrounds and specialized knowledge sets. Considering ChatGPT's propensity for generating inaccurate material, users have the option to establish a connection between the language model and their database. By doing so, they can contextualize the persona and enhance the correctness of the generated content.
In our project ChatGPT was augmented with a background of an ecology professor involved in fieldwork for our case study site , along with historical details about the Temple of Demeter. We designed 10 questions based on the knowledge base, encompassing topics such as history, significance, architectural style, and building materials (Table 1). In addition, various labels were assigned to objects within the virtual environment, facilitating interaction between the avatar and such objects.
Table 1: Ten Questions for students to learn about Temple of Demeter
| Question |
|---|
| Q1:When was the Temple of Demeter at Sangri built, and what materials were used in its construction? |
| Q2:What is the primary significance of the Temple of Demeter at Sangri? |
| Q3:Who was responsible for commissioning the construction of the Temple of Demeter? |
| Q4:Which deity was the Temple of Demeter dedicated to? |
| Q5:What unusual feature did the columns of the Temple of Demeter have? |
| Q6:What purpose did the two doors in the entrance area of the temple serve? |
| Q7:How was the roof of the Temple of Demeter constructed? |
| Q8:Which god was intended to be worshiped in the temple alongside Demeter? |
| Q9:What did the outer layer of the temple's walls consist of, and how were they linked? |
| Q10:Who conducted the investigations and excavations of the Temple of Demeter at Sangri? |
Figure 2 depicts a flowchart illustrating how an AI avatar engages in interactions with a user.
Fig. 2: Information architecture of VR environment with generative AI virtual avatar
Realistic 3D models, incorporating accurate materials and simulated lighting effects, serve as the foundational elements for digital exploration. Meanwhile, widely available Meta Oculus headset devices offer the necessary hardware support to enhance user immersion. These advancements have enabled the integration of AI avatars into virtual reality, allowing for conversational scene interactions that enhance user immersion.
Fig. 3: The digital model (Left) and photo (Right) of the Temple of Demeter, Naxos Island, Greece
This project involved the development of a user interaction framework within a VR environment. The participants were provided with a Meta Oculus controller to navigate and engage with the virtual scene. The fundamental operations encompass navigating within the environment, rotating one's perspective, engaging in dialogue with the AI avatar, manipulating objects within the area, posing inquiries using the inquiry control panel, requesting the AI avatar to accompany, and engaging in further forms of interaction.
The practical implementation entails the integration of genuine historical heritage site into a virtual environment, wherein the textures, terrain, and proportions are faithfully replicated to mirror their real-world counterparts. Furthermore, the AI-powered avatars are provided with narrative frameworks that are intricately constructed from the ecologist's personal life story and the historical narrative linked to the respective location. This synthesis offers users an exceptional experience by immersing them in a virtual environment and facilitating a meaningful and evidence-based dialogue with the created AI database. The present workflow, characterized by its inventive nature, has given rise to a diverse range of practical applications, all of which have the potential to bring about significant transformations. The primary use of this technology lies in the domain of immersive learning experiences. In the realm of education, students are bestowed with the ability to not only passively observe their surroundings, but also engage in active involvement, hence leading to the acquisition of knowledge through experiential learning. The integration of an interactive environment with AI-driven interactions enhances the learning paradigm by transforming the venue into an engaged pedagogical collaborator. This approach has a broader scope and appeals to various fields of study. The applications of this technology are extensive, as it enables the generation of computer models that accurately represent the physical environment.
Fig. 4: (a) Talk with AI virtual avatar; (b) Use teleport to navigate; (c) Question board for studying (see table 1)
The method used is a rapid and resource-efficient approach that offers several advantages. Firstly, it significantly decreases the time required for the modeling process. Additionally, it effectively maintains the authenticity of the materials and textures used in the model. Furthermore, it ensures the safeguarding of geographic and topographic data. This approach also presents limitations: it necessitates the utilization of a requiring high-performance computing systems to handle the processing and generation of photos and models, and intricate details, particularly those pertaining to plant components, may be challenging to capture.
After updating the database with the supplemental information responses provided by the AI avatar were evaluated for accuracy. The responses (Appendix 3), were factually accurate. and the language used by the avatar proved to be natural sounding. The synchronization observed in this context demonstrated a very engaging form of character interaction that effectively replicated the educational experience encountered during the field trip. Furthermore, this engagement extends beyond simple dialogue and influences the surrounding environment. Various labels are assigned to objects within the virtual area, facilitating interaction between the avatar and such objects. When a user engages with a designated destination, the virtual representation of the user, known as the avatar, spontaneously offers interpretive insights that augment the educational encounter.
Through the interaction of the intelligent virtual avatar with its knowledge content, background information, modeling of the 3D environment, material matching, light and shadow rendering, and physical system simulation, our research successfully integrated the generative AI avatar with the real-world 3D scene. The digital medium offers the potential to accurately replicate real-world landscapes and structures, thereby facilitating advancements in architectural and urban planning, virtual tourism, and historic preservation. The significance of applications within the realm of environmental protection cannot be overstated. The utilization of technology facilitates the opportunity for individuals to engage in immersive environmental awareness tours, allowing for virtual excursions and investigation of ecologically sensitive locations. The utilization of immersive interactions with the AI avatar aligns seamlessly with this objective, as it fulfills a dual role as an informative guide and a platform for engaging in discussions pertaining to sustainability. Furthermore, the proposed workflow enhances its transformative influence on the advancement of smart city development through the establishment of a digital twin. By employing geographic precision and advanced modeling techniques, the creation of digital replicas enables informed urban planning and decision-making processes.
Our digital temple of Demeter model is generated using a consumer-grade drone, DJI Mini 3 pro with 4K footage at 75% quality, so there is a lack of accuracy and detail in the model, which affects the user’s experience. Additionally, the scope of interaction options with our AI avatar is constrained. We have implemented only voice command interaction and question panels as means for users to engage with the system. Furthermore, user’s interactions with the virtual environment are restricted to roaming and object grabbing, without the AI avatar having the capability to access information within environment objects or facilitate more diverse interaction modalities. For the current research, we need to design experiments to validate the combination of generative AI avatars and point cloud modeling in terms of user immersion and immersive learning experience.
In the realm of digital replication and 3D modeling, our future endeavors encompass a comprehensive exploration of advanced techniques. Beyond the utilization of point cloud modeling, we aim to integrate algorithm-based neural rendering approaches, exemplified by Nvidia's Neurolangelo. These neural surface-based 3D reconstruction methods excel not only in preserving model geometry but also in faithfully representing material properties. Concurrently, Nvidia is actively harnessing AI-assisted point cloud modeling to iteratively refine models, progressively bridging the gap between virtual and real-world fidelity. In addition to using AI-enabled point-cloud modeling techniques, modeling using neural radiation fields (NeRF) is also our future direction of inquiry. These modeling techniques serve as vital underpinnings for elevating the realism of VR scenarios, facilitating swifter and more efficient digital world development for developers. Our forthcoming research agenda necessitates the systematic evaluation of diverse modeling techniques to complement and enhance the existing modeling workflow deployed in our current research. This evaluation will extend to a multifaceted approach, encompassing visual, auditory, and emotional recognition, eye tracking, and body posture analysis of AI avatars. These aspects hold significant promise for augmenting the user's experience within virtual environments, aligning with user-centered interaction design principles. Moreover, our research will entail training AI avatars to recognize virtual objects and designing interaction behaviors tailored to user movements and object interactions within the virtual environment. To validate the effectiveness of the design strategies, we will embark on a series of designed experiments, with a particular focus on assessing their impact on user immersion and immersive learning experiences. We envision that the workflow currently under investigation possesses broad applicability across various domains, and we invite fellow researchers to explore and validate its effectiveness within their respective areas of expertise.
In this paper, we present a workflow for VR immersive learning experience using point-cloud modeling and AI-driven avatars. By combining point-cloud modeling, generative AI, and VR technologies, our work not only achieves the digital modeling of architectural heritage site, but also provides digital tours and immersive learning experiences for the Temple of Demeter at Sangri, Greece.
The integration of generative AI, digital 3D models, and Unreal Engine 5 opens the way for transformative change. This research exemplifies the potential to recreate physical environments in the virtual realm, promote immersive education, and catalyze advancements in various fields. This paper provides a series of technical routes and design insights for using a combination of point-cloud modeling, generative AI, and VR technologies to support immersive teaching and learning experiences with the preservation and digitization of historic architectural heritage. With uncharted territories awaiting exploration, this innovative approach signifies the inception of a paradigm shift.
| Reference |
|---|
| Abramski, K., Citraro, S., Lombardi, L., Rossetti, G., & Stella, M. (2023). Cognitive Network Science Reveals Bias in GPT-3, GPT-3.5 Turbo, and GPT-4 Mirroring Math Anxiety in High-School Students. Big Data and Cognitive Computing, 7(3), Article 3. https://doi.org/10.3390/bdcc7030124 |
| Afrin, S., Rahman, A., Islam, F., & Hoque, F. (2013). ENVIRONMENTAL EFFECTS OF TOURISM. 1(7). |
| Aydın Ö., & Karaarslan E. (2023). Is ChatGPT Leading Generative AI? What is Beyond Expectations? (SSRN Scholarly Paper 4341500). https://doi.org/10.2139/ssrn.4341500 |
| Baidoo-Anu, D., & Owusu Ansah, L. (2023). Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning (SSRN Scholarly Paper 4337484). https://doi.org/10.2139/ssrn.4337484 |
| Biswas, S. (2023). The Function of chat GPT in Social Media: According to chat GPT. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.4405389 |
| Castiglioni, I., Rundo, L., Codari, M., Di Leo, G., Salvatore, C., Interlenghi, M., Gallivanone, F., Cozzi, A., D’Amico, N. C., & Sardanelli, F. (2021). AI applications to medical images: From machine learning to deep learning. Physica Medica, 83, 9–24. https://doi.org/10.1016/j.ejmp.2021.02.006 |
| Chheang, V., Marquez-Hernandez, R., Patel, M., Rajasekaran, D., Sharmin, S., Caulfield, G., Kiafar, B., Li, J., & Barmaki, R. L. (2023). Towards Anatomy Education with Generative AI-based Virtual Assistants in Immersive Virtual Reality Environments (arXiv:2306.17278). arXiv. https://doi.org/10.48550/arXiv.2306.17278 |
| Dwivedi, Y. K., Kshetri, N., Hughes, L., Slade, E. L., Jeyaraj, A., Kar, A. K., Baabdullah, A. M., Koohang, A., Raghavan, V., Ahuja, M., Albanna, H., Albashrawi, M. A., Al-Busaidi, A. S., Balakrishnan, J., Barlette, Y., Basu, S., Bose, I., Brooks, L., Buhalis, D., … Wright, R. (2023). Opinion Paper: “So what if ChatGPT wrote it?” Multidisciplinary perspectives on opportunities, challenges and implications of generative conversational AI for research, practice and policy. International Journal of Information Management, 71, 102642. https://doi.org/10.1016/j.ijinfomgt.2023.102642 |
| Evelpidou, N., Petropoulos, A., Karkani, A., & Saitis, G. (2021). Evidence of Coastal Changes in the West Coast of Naxos Island, Cyclades, Greece. Journal of Marine Science and Engineering, 9(12), Article 12. https://doi.org/10.3390/jmse9121427 |
| Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2014). Generative Adversarial Nets. Advances in Neural Information Processing Systems, 27. https://papers.nips.cc/paper_files/paper/2014/hash/5ca3e9b122f61f8f06494c97b1afccf3-Abstract.html |
| Grassini, S. (2023). Shaping the Future of Education: Exploring the Potential and Consequences of AI and ChatGPT in Educational Settings. Education Sciences, 13(7), 692. https://doi.org/10.3390/educsci13070692 |
| Haleem, A., Javaid, M., & Singh, R. P. (2022). An era of ChatGPT as a significant futuristic support tool: A study on features, abilities, and challenges. BenchCouncil Transactions on Benchmarks, Standards and Evaluations, 2(4), 100089. https://doi.org/10.1016/j.tbench.2023.100089 |
| Haskins, J., Zhu, B., Gainer, S., Huse, W., Eadara, S., Boyd, B., Laird, C., Farantatos, J., & Jerald, J. (2020). Haskins, J., Zhu, B., Gainer, S., Huse, W., Eadara, S., Boyd, B., ... & Jerald, J. (2020, March). Exploring VR training for first responders. In 2020 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW) (pp. 57-62). IEEE. 2020 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW), 57–62. https://doi.org/10.1109/VRW50115.2020.00018 |
| Jeon, J., Lee, S., & Choe, H. (2023). Beyond ChatGPT: A conceptual framework and systematic review of speech-recognition chatbots for language learning. Computers & Education, 206, 104898. https://doi.org/10.1016/j.compedu.2023.104898 |
| Kiourt, C., Pavlidis, G., Koutsoudis, A., & Kalles, D. (2017). Realistic Simulation of Cultural Heritage. International Journal of Computational Methods in Heritage Science (IJCMHS), 1(1), 10–40. https://doi.org/10.4018/IJCMHS.2017010102 |
| Ku, H., & Lee, M. (2023). TextControlGAN: Text-to-Image Synthesis with Controllable Generative Adversarial Networks. Applied Sciences, 13(8), Article 8. https://doi.org/10.3390/app13085098 |
| Leka, A., Lagarias, A., Panagiotopoulou, M., & Stratigea, A. (2022). Development of a Tourism Carrying Capacity Index (TCCI) for sustainable management of coastal areas in Mediterranean islands – Case study Naxos, Greece. Ocean & Coastal Management, 216, 105978. https://doi.org/10.1016/j.ocecoaman.2021.105978 |
| LI, C., Xu, T., Zhu, J., & Zhang, B. (2017). Triple Generative Adversarial Nets. Advances in Neural Information Processing Systems, 30. https://proceedings.neurips.cc/paper_files/paper/2017/hash/86e78499eeb33fb9cac16b7555b50767-Abstract.html |
| Pepe, M., Alfio, V. S., Costantino, D., & Scaringi, D. (2022). Data for 3D reconstruction and point cloud classification using machine learning in cultural heritage environment. Data in Brief, 42, 108250. https://doi.org/10.1016/j.dib.2022.108250 |
| Konar, A. (2018). Artificial intelligence and soft computing: behavioral and cognitive modeling of the human brain. CRC press. |
| Radford, A., Metz, L., & Chintala, S. (2015, November 19). Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks. arXiv.Org. https://arxiv.org/abs/1511.06434v2 |
| Radianti, J., Majchrzak, T. A., Fromm, J., & Wohlgenannt, I. (2020). A systematic review of immersive virtual reality applications for higher education: Design elements, lessons learned, and research agenda. Computers & Education, 147, 103778. https://doi.org/10.1016/j.compedu.2019.103778 |
| Roumana, A., Georgopoulos, A., & Koutsoudis, A. (2022). Dwivedi. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XLIII-B2-2022, 885–891. https://doi.org/10.5194/isprs-archives-XLIII-B2-2022-885-2022 |
| Shen, T., Munkberg, J., Hasselgren, J., Yin, K., Wang, Z., Chen, W., Gojcic, Z., Fidler, S., Sharp, N., & Gao, J. (2023). Flexible Isosurface Extraction for Gradient-Based Mesh Optimization. ACM 图形交易, 42(4), 37:1-37:16. https://doi.org/10.1145/3592430 |
| Walmsley, A. P., & Kersten, T. P. (2020). The Imperial Cathedral in Königslutter (Germany) as an Immersive Experience in Virtual Reality with Integrated 360° Panoramic Photography. Applied Sciences, 10(4), 1517. https://doi.org/10.3390/app10041517 |
Nikos Katsaros is a native Greek, born and raised on the beautiful island of Naxos in the Aegean Sea. Growing up amidst the stunning landscapes and unique biodiversity of the island, Nikos developed a deep appreciation for nature from an early age. He was always fascinated by the interconnectedness of the ecosystem and the delicate balance that sustains life on Naxos. As he matured, Nikos' passion for the environment grew stronger, leading him to pursue a career as an ecologist. He left Naxos to study environmental science at a prominent university in Athens, where he further honed his knowledge and skills in understanding and protecting natural ecosystems. After completing his education, Nikos felt a strong calling to return to his homeland and contribute to the conservation efforts of Naxos. He believed that the island's unique biodiversity and delicate ecosystems deserved protection, especially in the face of increasing human activities and global environmental challenges. Back on Naxos, Nikos immersed himself in the study and preservation of the island's diverse habitats, which range from lush green valleys to rugged mountains and pristine beaches. He worked closely with local environmental organizations and government authorities to raise awareness about the importance of sustainability and the need to protect the island's natural heritage. Nikos also collaborated with local communities, engaging them in environmental initiatives and empowering them to adopt eco-friendly practices. His passion and knowledge were evident in the workshops and educational programs he organized to instill a sense of pride and responsibility for Naxos' environment in the hearts of its residents. As an ecologist, Nikos constantly explored and researched the island's ecosystems, studying the flora and fauna unique to Naxos. He was an advocate for sustainable tourism, advocating for responsible travel that minimized negative impacts on the environment while still allowing people to appreciate and enjoy the island's natural beauty. In his spare time, Nikos enjoyed hiking through the island's diverse landscapes, birdwatching, and even scuba diving to explore the thriving underwater world surrounding Naxos. His experiences in nature further fueled his commitment to conservation, and he was always eager to share his knowledge with anyone who expressed an interest in protecting the environment. Nikos Katsaros' unwavering dedication to preserving Naxos' ecological treasures earned him respect and admiration among the local community and fellow environmentalists alike. He remained a steadfast guardian of the island's natural wonders, always striving to leave a lasting positive impact on the place he loved and called home.
Temple of Demeter at Sangri, built completely from Naxian marble around 530 BC. Dedicated to Demeter, the Greek goddess of agriculture, harvest, and fertility, it is no surprise that its location has views of lush, fertile farmland for kilometers. Interestingly, this temple contains several features that are unique when compared to other architecture from this period, including a South facing facade instead of the usual East facing ones, and the columns taper towards the top, whereas classical Greek columns taper towards the bottom. For these reasons, the temple is considered of great archeological importance and underwent extensive restoration in the mid-1990s. Parts of the temple not used in the restoration are displayed in the small archeological museum, where visitors can also find digital renderings and other in-depth details about the original temple architecture.
Near the village of Sangri, 10 km from Naxos Town, lies the sacred temple of Demeter. This magnificent temple has been made exclusively of the finest quality Naxos marble and dates back to the 6th century BC.
As Demeter was the Olympian goddess of agriculture, harvest, and fertility, people built temples dedicated to her on fertile lands, such as that of Sangri.
Excavations have led to the conclusion that the area was serving religious purposes since the late Mycenaean Era. During that time, places of worship used to be open-air.
The temple of Demeter in its current form was built around 530-520 BC, in the Doric architectural style that later influenced the Parthenon of Athens. It is of major archaeological importance, and some of its best-preserved sections have helped experts gain a better understanding of the architecture of that period.
Its construction took place under the rule of Lygdamis, the tyrant of Naxos. Lygdamis had planned to create some of Greece's most impressive monuments in Naxos, including the Temple of Demeter and Portara, which supposedly would be a part of a great temple dedicated to Apollo that was never completed.
The Temple of Sangri is a Late Archaic Greek temple on the Cycladic island of Naxos in the area of Gyroulas, about 1.5 km south of Ano Sangri [de]. The temple was built around 530 BC and is one of the earliest Ionic temples. It was built completely from Naxian marble.
The temple was built around 530 BC. Based on the finds, the sanctuary was probably dedicated to Demeter or perhaps Kore. For this reason and because of its unusual shape, the temple is often referred to as a telesterion. There are also indications of a cult to Apollo on the site. If still in use by the 4th century AD, the temple would have been closed during the persecution of pagans under the Christian emperors. In the 6th century AD, the temple was largely demolished and a three-naved Christian basilica was built from the same stone on the same site.
The temple contains many unusual features. The ground plan is almost square (13.29 x 12.73 m), when Greek temples, especially in the archaic period are usually elongated. The facade was on the south side instead of the usual location on the east or (much more rarely) west side. The temple was built without a foundation platform (the Crepidoma), directly on top of the euthynteria, likewise there is no stylobate for the columns.
The facade was formed of five columns in antis. The columns are in the Samian style, but unfluted. Very unusually, the columns show a slight contraction in width with increasing height, whereas Archaic and Classical Greek columns as a rule increase in thickness towards the top (entasis). The leafy wreathed capitals were not carved in the round, but were instead painted onto the double echinus of the capitals, while the abacus was decorated with a band.
On the smooth architrave, hidden behind plates on the facade, were the ends of roof beams which supported the oldest known marble roof of ancient architecture, which covered the pronaos. Seven almost four meter long beams formed the purlins of this roof. The beams were curved upwards almost 2 cm, which affected all the structural elements of the roof above them, giving the roof a slight curvature.
Two doors, placed in line with the second and fifth intercolumniation, led from the entrance area to the naos. The doorway was decorated with bands lined with a large bead and reel pattern. Above the door was a painted cymatium, which supported a smooth undecorated lintel. The naos was divided into two parts by a row of five columns which were in line with the columns of the facade. The columns in the centre of the room rise to a height of 5.4 to 6.46 metres, but despite the different heights they all have the same diameter at ground level (50 cm) – in violation of the norms of archaic rules of proportion. They do not narrow towards the top. Like the facade columns they stand atop smooth Samian bases, but these have no torus. These columns supported the marble struts of a saddle roof, which ran from front to back about 4 metres in each direction. Diffuse light would have entered the naos through the gaps in the roof tiles even when the doors to the naos were closed.
The walls of the naos rested on a toichobate which was about 28 cm high and 70 cm wide and composed of two layers. While the outer layer was made of square blocks, the inner layer was made from smaller stones of irregular shape, giving it an unusual "rough" appearance. The space between these two layers was filled with rubble and marble splinters. The blocks of the outer layer show anathyrosis on the short sides, but not on the smooth long sides. The blocks were only occasionally linked with wooden dowels, mostly near the cornice. The outer layer was inclined outwards at an angle of about 3 %, which, like the shape of the columns of the facade, was the opposite of the archaic norm, which was usually intended to ensure in inward inclination. The whole exterior was plastered and painted.
It is impressive that the whole building, which was built entirely from marble, is completely permeated by a common curvature, for which each individual part of the building had to be specially formed. This careful shaping of individual parts made the task of reconstruction easy, since the exact measurements of a large number of the 1600 or so surviving parts of the building indicated their original location.
Nikolaos Kondoleon investigated the temple in 1949. It was excavated and studied from 1976 to 1985 by Vassilis Lambrinoudakis and Gottfried Grubenas part of a partnership between the University of Athens and the Technical University of Munich. This culminated in a partial restoration of the structure and the construction of a small museum on the site, which was opened in August 2001. A final publication of the excavations is in progress.
| Q&A 1 |
|---|
| Q1: When was the Temple of Demeter at Sangri built, and what materials were used in its construction? |
| a) Around 530-520 BC, built entirely from Naxian marble. |
| b) Around 500-490 BC, built from limestone and sandstone. |
| c) Around 510-500 BC, built entirely from Parian marble. |
| d) Around 520-510 BC, built from limestone and granite. |
| A1 (Convai provided) : The Temple of Demeter at Sangri was built around 530 BC and was made completely of Naxian marble. It is considered of great archeological importance and underwent extensive restoration in the mid-1990s. Parts of the temple not used in the restoration are displayed in the small archeological museum. |
| Q&A 2 |
|---|
| Q2: What is the primary significance of the Temple of Demeter at Sangri? |
| a) It was the largest temple in ancient Greece. |
| b) It was a center of worship for various Greek gods and goddesses. |
| c) It showcased the earliest use of Corinthian columns in Greek architecture. |
| d) It is of major archaeological importance and provides insights into the architecture of its time. |
| A2 (Convai provided) : The primary significance of the Temple of Demeter at Sangri is its unique architecture and features, which distinguish it from other temples from this period. It is the only temple from that era to have a south facing facade, and its columns taper towards the top instead of the usual taper towards the bottom. Additionally, the temple has helped experts gain a better understanding of the architecture of that period. |
| Q&A 3 |
|---|
| Q3: Who was responsible for commissioning the construction of the Temple of Demeter? |
| a) Lygdamis, the tyrant of Naxos. |
| b) Pericles, the Athenian statesman. |
| c) Alexander the Great, the Macedonian conqueror. |
| d) King Leonidas, the Spartan warrior. |
| A3 (Convai provided) : The Temple of Demeter was commissioned by Lygdamis, the tyrant of Naxos, around 530-520 BC. |
| Q&A 4 |
|---|
| Q4: Which deity was the Temple of Demeter dedicated to? |
| a) Hera, the queen of the gods. |
| b) Zeus, the king of the gods. |
| c) Athena, the goddess of wisdom. |
| d) Demeter, the goddess of agriculture and fertility. |
| A4 (Convai provided) : The Temple of Demeter was dedicated to Demeter, the Greek goddess of agriculture, harvest, and fertility. |
| Q&A 5 |
|---|
| Q5: What unusual feature did the columns of the Temple of Demeter have? |
| a) They were fluted, following the typical Greek style. |
| b) They were entirely unadorned, without any capitals. |
| c) They tapered towards the bottom, like classical Greek columns. |
| d) They showed a slight contraction in width with increasing height. |
| A5 (Convai provided) : The columns of the Temple of Demeter had an unusual feature in that they taper towards the top, whereas classical Greek columns taper towards the bottom. |
| Q&A 6 |
|---|
| Q6: What purpose did the two doors in the entrance area of the temple serve? |
| a) They allowed access to the inner sanctuary. |
| b) They were purely decorative and had no functional use. |
| c) They were used for ventilation during religious ceremonies. |
| d) They were openings for the temple's guards to keep watch. |
| A6 (Convai provided) : The two doors in the entrance area of the temple served as entryways into the naos, which was the inner chamber of the temple. This chamber was separated from the entrance area by a row of five columns. The doors allowed access to the naos and also allowed diffuse light to enter the chamber, even when the doors were closed. |
| Q&A 7 |
|---|
| Q7: How was the roof of the Temple of Demeter constructed? |
| a) It had a flat roof made of wooden beams. |
| b) It featured a pyramidal roof made of clay tiles. |
| c) It had a saddle roof supported by marble struts. |
| d) It was a dome-shaped roof made of bronze. |
| A7 (Convai provided) : The roof of the Temple of Demeter was constructed using seven almost four meter long beams that formed the purlins of the roof. The beams were curved upwards almost 2 cm, which affected all the structural elements of the roof above them, giving the roof a slight curvature. The roof was covered with marble tiles and supported by columns in the naos. |
| Q&A 8 |
|---|
| Q8: Which god was intended to be worshiped in the temple alongside Demeter? |
| a) Apollo, the god of music and prophecy. |
| b) Poseidon, the god of the sea and earthquakes. |
| c) Hermes, the messenger god and patron of travelers. |
| d) Ares, the god of war and battle. |
| A8 (Convai provided) : There are indications of a cult to Apollo on the site, and it is thought that the Temple of Demeter was meant to be part of a larger temple dedicated to Apollo, although this was never completed. |
| Q&A 9 |
|---|
| Q9: What did the outer layer of the temple's walls consist of, and how were they linked? |
| a) Square blocks, linked with iron dowels. |
| b) Smaller stones of irregular shape, linked with wooden dowels. |
| c) Smooth marble slabs, linked with copper dowels. |
| d) Rough limestone blocks, linked with stone mortar. |
| A9 (Convai provided) : The outer layer of the temple walls consisted of square blocks that showed anathyrosis on the short sides, but not on the smooth long sides. The blocks were only occasionally linked with wooden dowels, mostly near the cornice. The space between the two layers was filled with rubble and marble splinters. |
| Q&A 10 |
|---|
| Q10: Who conducted the investigations and excavations of the Temple of Demeter at Sangri? |
| a) Nikolaos Kondoleon in 1949. |
| b) Vassilis Lambrinoudakis and Gottfried Gruben from 1976 to 1985. |
| c) A team of French archaeologists in the 1960s. |
| d) A group of local volunteers in the 2000s. |
| A10 (Convai provided) : Nikolaos Kondoleon investigated the temple in 1949. It was excavated and studied from 1976 to 1985 by Vassilis Lambrinoudakis and Gottfried Grubenas part of a partnership between the University of Athens and the Technical University of Munich. This culminated in a partial restoration of the structure and the construction of a small museum on the site, which was opened in August 2001. A final publication of the excavations is in progress. |